icuSources/tools/genrb/rle.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 *
   6 *   Copyright (C) 2000, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 *******************************************************************************
  10 *
  11 * File writejava.c
  12 *
  13 * Modification History:
  14 *
  15 *   Date        Name        Description
  16 *   01/11/02    Ram        Creation.
  17 *******************************************************************************
  18 */
  19
  20 #ifndef RLE_H
  21 #define RLE_H 1
  22
  23 #include "unicode/utypes.h"
  24 #include "unicode/ustring.h"
  25
  26 U_CDECL_BEGIN
  27 /**
  28  * Construct a string representing a byte array.  Use run-length encoding.
  29  * Two bytes are packed into a single char, with a single extra zero byte at
  30  * the end if needed.  A byte represents itself, unless it is the
  31  * ESCAPE_BYTE.  Then the following notations are possible:
  32  *   ESCAPE_BYTE ESCAPE_BYTE   ESCAPE_BYTE literal
  33  *   ESCAPE_BYTE n b           n instances of byte b
  34  * Since an encoded run occupies 3 bytes, we only encode runs of 4 or
  35  * more bytes.  Thus we have n > 0 and n != ESCAPE_BYTE and n <= 0xFF.
  36  * If we encounter a run where n == ESCAPE_BYTE, we represent this as:
  37  *   b ESCAPE_BYTE n-1 b
  38  * The ESCAPE_BYTE value is chosen so as not to collide with commonly
  39  * seen values.
  40  */
  41 int32_t
  42 byteArrayToRLEString(const uint8_t* src,int32_t srcLen, uint16_t* buffer,int32_t bufLen, UErrorCode* status);
  43
  44
  45 /**
  46  * Construct a string representing a char array.  Use run-length encoding.
  47  * A character represents itself, unless it is the ESCAPE character.  Then
  48  * the following notations are possible:
  49  *   ESCAPE ESCAPE   ESCAPE literal
  50  *   ESCAPE n c      n instances of character c
  51  * Since an encoded run occupies 3 characters, we only encode runs of 4 or
  52  * more characters.  Thus we have n > 0 and n != ESCAPE and n <= 0xFFFF.
  53  * If we encounter a run where n == ESCAPE, we represent this as:
  54  *   c ESCAPE n-1 c
  55  * The ESCAPE value is chosen so as not to collide with commonly
  56  * seen values.
  57  */
  58 int32_t
  59 usArrayToRLEString(const uint16_t* src,int32_t srcLen,uint16_t* buffer, int32_t bufLen,UErrorCode* status);
  60
  61 /**
  62  * Construct an array of bytes from a run-length encoded string.
  63  */
  64 int32_t
  65 rleStringToByteArray(uint16_t* src, int32_t srcLen, uint8_t* target, int32_t tgtLen, UErrorCode* status);
  66 /**
  67  * Construct an array of shorts from a run-length encoded string.
  68  */
  69 int32_t
  70 rleStringToUCharArray(uint16_t* src, int32_t srcLen, uint16_t* target, int32_t tgtLen, UErrorCode* status);
  71
  72 U_CDECL_END
  73
  74 #endif